From a86787e873551425723771a6aa1e69013f6a183d Mon Sep 17 00:00:00 2001
From: Stefan Heng <stefan.hg@outlook.com>
Date: Mon, 22 May 2023 12:20:28 -0400
Subject: [PATCH] Chore: publish; Fix: gpt3 auth token

---
 setup.py                           |  4 +-
 zeroshot_classifier/models/gpt3.py | 65 +++++++++++++++++-------------
 2 files changed, 39 insertions(+), 30 deletions(-)

diff --git a/setup.py b/setup.py
index 79efdd5..95beab4 100644
--- a/setup.py
+++ b/setup.py
@@ -1,6 +1,6 @@
 from setuptools import setup, find_packages
 
-VERSION = '0.2.1'
+VERSION = '0.2.2'
 DESCRIPTION = """
 code and data for the Findings of ACL'23 paper Label Agnostic Pre-training for Zero-shot Text Classification 
 by Christopher Clarke, Yuzhao Heng, Yiping Kang, Krisztian Flautner, Lingjia Tang and Jason Mars
@@ -15,7 +15,7 @@
     description=DESCRIPTION,
     long_description=DESCRIPTION,
     url='https://github.com/ChrisIsKing/zero-shot-text-classification',
-    download_url='https://github.com/ChrisIsKing/zero-shot-text-classification/archive/refs/tags/v0.2.1.tar.gz',
+    download_url='https://github.com/ChrisIsKing/zero-shot-text-classification/archive/refs/tags/v0.2.2.tar.gz',
     packages=find_packages(),
     include_package_data=True,
     install_requires=[
diff --git a/zeroshot_classifier/models/gpt3.py b/zeroshot_classifier/models/gpt3.py
index 92e681b..00411c4 100644
--- a/zeroshot_classifier/models/gpt3.py
+++ b/zeroshot_classifier/models/gpt3.py
@@ -36,22 +36,27 @@ class ApiCaller:
     """
     url = 'https://api.openai.com/v1/completions'
 
-    with open(os_join(u.proj_path, 'auth', 'open-ai.json')) as f:
-        auth = json.load(f)
-        # api_key, org = auth['api-key'], auth['organization']
-        api_key, org = auth['api-key-chris'], auth['organization']
-    headers = {
-        'Content-Type': 'application/json',
-        'Authorization': f'Bearer {api_key}',
-        'OpenAI-Organization': org
-    }
-
-    def __init__(self, model: str = 'text-ada-001', batched: bool = False, delay: float = None):
+    def __init__(
+            self, model: str = 'text-ada-001', batched: bool = False, delay: float = None,
+            token_path: str = None
+    ):
         self.model = model
         self.batched = batched
 
         self.delay = delay
 
+        token_path = token_path or os_join(u.proj_path, 'auth', 'open-ai.json')
+        assert os.path.exists(token_path), f'OpenAI token not found at {pl.i(token_path)}'
+        with open(token_path) as f:
+            auth = json.load(f)
+            api_key, org = auth['api-key'], auth['organization']
+            openai.api_key = api_key
+        # self.headers = {
+        #     'Content-Type': 'application/json',
+        #     'Authorization': f'Bearer {api_key}',
+        #     'OpenAI-Organization': org
+        # }
+
     @retry(wait=wait_random_exponential(min=1, max=60 * 30))  # Wait for 30min
     def completion(self, **kwargs):
         if self.delay:
@@ -413,11 +418,13 @@ def parse_args():
 if __name__ == '__main__':
     mic.output_width = 256
 
-    with open(os_join(u.proj_path, 'auth', 'open-ai.json')) as f:
-        auth = json.load(f)
-        org = auth['organization']
-        api_key = auth['api-key']
-    openai.api_key = api_key
+    def set_openai_token():
+        with open(os_join(u.proj_path, 'auth', 'open-ai.json')) as f:
+            auth = json.load(f)
+            org = auth['organization']
+            api_key = auth['api-key']
+        openai.api_key = api_key
+    # set_openai_token()
 
     # evaluate(model='text-ada-001', domain='in', dataset_name='emotion')
     # evaluate(model='text-curie-001', domain='out', dataset_name='multi_eurlex', concurrent=True)
@@ -429,18 +436,20 @@ def parse_args():
     # evaluate(model='text-davinci-002', domain='out', dataset_name='consumer_finance')
     # evaluate(model='text-davinci-002', domain='out', dataset_name='amazon_polarity', concurrent=True)
 
-    run_args = dict(model='text-curie-001', subsample=True, store_meta=True, store_frequency=10)
-    # dnm = 'amazon_polarity'
-    # dnm = 'yelp'
-    # dnm_ = 'consumer_finance'
-    # dnm_ = 'slurp'
-    dnm_ = 'multi_eurlex'  # TODO: doesn't work w/ batched requests???
-    # dnm_ = 'sgd'
-    rsm = [os_join(
-        u.eval_path, '2022-12-04_17-34-01_Zeroshot-GPT3-Eval_{md=text-curie-001, dm=out, dnm=multi_eurlex}',
-        '22-12-04_out-of-domain', f'{dnm_}_meta.json'
-    )]
-    # evaluate(domain='out', dataset_name=dnm_, **run_args, concurrent=True, delay=12, resume=rsm)
+    def subsample_large_dset():
+        run_args = dict(model='text-curie-001', subsample=True, store_meta=True, store_frequency=10)
+        # dnm = 'amazon_polarity'
+        # dnm = 'yelp'
+        # dnm_ = 'consumer_finance'
+        # dnm_ = 'slurp'
+        dnm_ = 'multi_eurlex'  # TODO: doesn't work w/ batched requests???
+        # dnm_ = 'sgd'
+        rsm = [os_join(
+            u.eval_path, '2022-12-04_17-34-01_Zeroshot-GPT3-Eval_{md=text-curie-001, dm=out, dnm=multi_eurlex}',
+            '22-12-04_out-of-domain', f'{dnm_}_meta.json'
+        )]
+        evaluate(domain='out', dataset_name=dnm_, **run_args, concurrent=True, delay=12, resume=rsm)
+    # subsample_large_dset()
 
     def command_prompt():
         args = parse_args()